天津市近海水质预测方法研究  

Prediction method for Tianjin coastal water quality

作　　者：牛志广[1] 陆仁强[1] 王晨婉[1] 

机构地区：[1]天津大学环境科学与工程学院,天津300072

出　　处：《水科学进展》2009年第2期222-226,共5页Advances in Water Science

基　　金：天津市自然科学基金资助项目(07JCYBJC07200)~~

摘　　要：针对中国近海水质污染呈现的年周期性和自相似性特点,通过对分形理论的研究,提出了一种新的完全基于环境监测数据的近海水质污染变化的分形预测方法。以天津市近岸海域为例,根据分形拼贴定理,由基于仿射变换的分形插值方法求取各历史时间阶段水质变化的迭代函数系,根据近海水质变化的年周期性,对上述求得的迭代函数系加权求和,得到预测年份水质变化的统计意义上的迭代函数系,建立分形预测模型,应用随机迭代算法求得预测年份水质变化曲线的吸引子,对近海水质进行预测,预测结果显示枯水期、丰水期、平水期的平均预测误差分别为29.6%、27.5%、16.1%,3个水期的平均预测误差为24.4%。应用表明,该方法预测精度较高、实用性强,能够为近海水环境管理提供决策支持。According to the periodicity and self-similarity, a new method for predicting the change characteristics of the coastal water quality is proposed through the study of the fractional theory based on the environmental monitoring data completely. Taking the coastal marine of Tianjin as an example, firstly, according to the fractional collage theory, the fractal interpolation based on affine transform is used to find the iterated function systems of the historical water quality. Secondly, the weighted summation method is adopted to find the iterated function system of the predicting period according to the above iterated function systems based on the periodicity. Then the fractal predicting model is established according to the iterated function system of predicting period. Finally, the random iterated algorithm is used to find the attractor of each prediction periods which can provide the prediction data according to the time values. The predicted results show that the average prediction error of the dry, rainy and mean water period are 29.6%, 27.5% and 16.1%, respectively, and the average prediction error of the three periods is 24.4 %. Through application it can be found that the fractional prediction method has high prediction preeiseon. It is practicable and can povide the decision-making for the environment management of the coastal marine.

关 键 词：近海水质 预测 分形理论 拼贴定理 迭代函数系 

分 类 号：X55[环境科学与工程—环境工程]